鶹ý is launching a series investigating complications from endometrial ablation. Here we explain our methods for analyzing FDA adverse event reports related to NovaSure.
Attuned to the limitations of FDA Manufacturer and User Facility Device Experience (MAUDE) data, the 鶹ý investigative team was conservative in its evaluation of reports appearing in the database.
Team members searched on the brand name NovaSure for the full years from 2009 through 2019. (Reports for 2009 were removed on Jan. 1, 2020, as the public database is trimmed annually to reflect only the previous 10 years.)
A member of the team read through each of 1,815 reports and logged it, if it qualified, into one of more of 10 categories as appropriate: death, cardiac arrest/bradycardia, sepsis, infection, hysterectomy, bowel surgery, burns, perforations, hematometra, or post-ablation tubal sterilization syndrome. Reports solely about device malfunctions or other minor complaints were discarded. Multiple complications mentioned in a single report were logged separately, such that the total number of complications identified in 鶹ý's analysis outnumbered the individual MAUDE reports that were included.
Reports also were discounted if they were unclear about the NovaSure device's potential role, such as: "A hysteroscopy was performed prior to the attempted ablation, as was a dilatation and sounding with a metal sound (not a Hologic device). It is not known when this perforation occurred or what instrument may have been the cause."
Thus, numbers of complications would be higher if those reports were included. Even though MAUDE is already known to be under-reported, we preferred to err on the side of caution.
Another known limitation of MAUDE is that individual reports are not adjudicated. 鶹ý did not confirm each report individually.
Finally, a lack of reliable denominator data -- that is, the total number of NovaSure procedures performed during our review period -- prevented us from calculating complication rates.